{"id":"54063ae6-c947-4989-8401-0735d3102b7d","arxiv_id":"2507.06995","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"For special Gushel-Mukai threefolds, the invariant part of the infinitesimal period map is injective, and the kernel of the full period map has dimension exactly three.","lead":"This paper proves that the invariant part of the infinitesimal period map is injective for special Gushel-Mukai threefolds, a double-cover analogue of the hyperelliptic curve case. It also computes the kernel of the infinitesimal period map for other Fano threefolds and interprets these kernels using Bridgeland moduli spaces in Kuznetsov components.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The Hodge proof of the main injectivity result hinges on the unproved commutativity lemma (Lemma 2.17); if the diagram is not verified, the reduction to Proposition 2.16 collapses.","rationale":"I read Lemma 2.19 as likely correct: the Serre duality and Kodaira-Akizuki-Nakano vanishings check out, and the conclusion is consistent with h^{1,1}(Y)=1. Lemma 2.8's sign argument is indeed inconsistent, but the anti-invariant part is zero by Proposition 2.2 and Remark 2.3, so that flaw does not threaten the theorem. The categorical proof rests on a chain of published results; if those references are accurate, it independently supports the main claim. The single most load-bearing internal gap is therefore the unproved Lemma 2.17, which is exactly the point where the Hodge-theoretic reduction could fail. Since this is a missing derivation rather than a detected falsehood, the appropriate disposition is the reader's conditional verdict: the authors should supply a complete proof of Lemma 2.17 and clarify the surjectivity of the second vertical map.","tokens_in":548,"tokens_out":17291,"duration_ms":206326,"concrete_test":"Write out a complete proof of Lemma 2.17 from exact sequences (19) and (20) and [Fle86, Lemma 2.10], specifying every arrow and the sign. In particular, verify that for every nonzero tau in H^1(S,T_S)_0 and every beta in H^1(S,Omega^1_S(-1)), the diagram relates the top pairing m_1(tau,beta) to the bottom pairing m_2(I(tau),gamma) for the prescribed gamma, and check whether the vertical map on the second factor is surjective or otherwise enough to transfer non-degeneracy. If the diagram fails, or if the vertical map is not surjective without a substitute argument, Proposition 2.16 and Theorem 2.10 do not follow from the written Hodge proof.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The proof of Theorem 2.10 is reduced, via Lemma 2.15, to Proposition 2.16, which asserts non-degeneracy of the pairing on H^1(S,T_S)_0. The only bridge between that pairing and the pair of statements 'I is injective' and 'the bottom pairing is non-degenerate' is Lemma 2.17, a commutative diagram claimed up to sign. In the manuscript Lemma 2.17 is not proved: the text says only that it follows from exact sequences (19), (20) and [Fle86, Lemma 2.10]. The vertical maps are composites of contraction, residue, and boundary maps, so a sign or twist error would invalidate the reduction: even if Lemma 2.18 and Lemma 2.21 are correct, they would not imply Proposition 2.16. The reader's flagged assumption, Lemma 2.19, is a component of Lemma 2.18 and appears likely correct; the genuinely load-bearing internal gap is the unproved Lemma 2.17. The categorical proof is independent, but the paper advertises a Hodge-theoretic proof, and that proof is incomplete at this exact step.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies infinitesimal Torelli problems for special Gushel-Mukai threefolds and related Fano threefolds. The main theorem (Theorem 2.9/2.10) asserts that for a special Gushel-Mukai threefold X, the invariant part of the infinitesimal period map is injective, so that the kernel of the full infinitesimal period map is exactly the anti-invariant summand H^1(Y,T_Y(-1)) and has dimension 3. The proof is attempted from two perspectives: a Hodge-theoretic reduction to a twisted Torelli statement for the branch K3 surface S (Proposition 2.16), and a categorical argument via the Kuznetsov component and normal Hochschild cohomology. The paper also computes kernels for prime Fano threefolds of genus 7, 8, 9, 10, and 12, and gives a geometric interpretation of the kernel via Bridgeland moduli spaces.","tokens_in":29708,"tokens_out":10133,"duration_ms":99175,"significance":"If correct, the main result establishes the expected 'hyperelliptic-like' failure of the infinitesimal Torelli theorem for special Gushel-Mukai threefolds and provides a clean description of the kernel of the period map. The reduction to a K3-surface statement is a natural and potentially reusable technique, and the independent categorical proof, if complete, would be a valuable bridge between Hodge theory and derived categories. The paper is careful in many cohomological computations and makes good use of existing results on Gushel-Mukai threefolds and Kuznetsov components. However, the Hodge-theoretic proof currently has a load-bearing unproved lemma, and the categorical proof has an insufficiently justified spectral sequence degeneration, so the paper is not yet in final form.","major_comments":[{"comment":"Lemma 2.17 is stated without proof. The text only says that it follows from the exact sequences (19), (20) and [Fle86, Lemma 2.10], but no actual verification is provided. This lemma is the only bridge between the pairing on H^1(S,T_S)_0 × H^1(S,Ω^1_S(-1)) and the pair of statements (injectivity of I and non-degeneracy of the bottom row) that are used to prove Proposition 2.16. Since the vertical maps are composites of contraction, residue, and boundary maps, a sign or twist error would invalidate the reduction. Please supply a complete proof of the commutativity up to sign, including an explicit description of the maps and the sign convention, or give a precise reference that implies this exact diagram.","section":"§2.4, Lemma 2.17"},{"comment":"In the proof of Theorem 3.2(1), after computing the E_1-term of the normal Hochschild spectral sequence, the text asserts: 'It is also easy to see this normal Hochschild spectral sequence degenerates at the E_2-page.' This degeneration is load-bearing for the identification NHH^2(⟨O_X,U^∨_X⟩,X) ≅ Hom(U_X,Q^∨_X). Please provide the differential analysis (or a precise reference) showing that all relevant higher differentials vanish, since without this the categorical computation of Ker η is incomplete.","section":"§3.2, Theorem 3.2"}],"minor_comments":[{"comment":"The sign computation in the proof that the anti-invariant subspace lies in the kernel is incorrect: for β anti-invariant and α anti-invariant, one has ι(β·α) = β·α, not -β·α. The conclusion is nevertheless correct and follows from the block decomposition in Proposition 2.2 together with H^1(Y,Ω^2_Y) = H^2(Y,Ω^1_Y) = 0; please rewrite this part accordingly.","section":"§2.2, Lemma 2.8"},{"comment":"There are several typographical errors: 'infnitesimal' appears in the abstract and in Section 1, 'Grassmiann' appears in Lemma 2.14, 'Kunnenth' appears in Appendix A, 'Propostion A.2' appears in the Introduction, and 'replies' appears in Section 1.1 (should be 'relies'). These should be corrected.","section":"Throughout"},{"comment":"The cross-reference '(see Remark 2.6)' in the proof of Lemma 2.8 is likely intended to point to Remark 2.3, since Remark 2.6 concerns H^1(Y,T_Y(-log S)) rather than the invariant/anti-invariant decomposition.","section":"§2.2, proof of Lemma 2.8"},{"comment":"The notation for the moduli space of semistable sheaves is inconsistent: the statement of Theorem 3.5(2) uses M_X(2,0,4), while the proof and surrounding text use M^ss_X(2,0,4). Please make the notation uniform.","section":"§3, Theorem 3.5"},{"comment":"In the commutative diagram of Proposition 2.11, one vertical arrow is labeled only with a question mark. Please label it explicitly (as the inclusion H^1(Y,T_Y(-log S)) → H^1(S,T_S)) for clarity.","section":"§2.3, Proposition 2.11"}],"recommendation":"major_revision","confidential_remarks":"The paper's main conclusion is likely correct, and the categorical proof may be sufficient on its own. However, the Hodge-theoretic proof advertised in the title and introduction has a clear gap at Lemma 2.17, and the categorical proof also needs a documented spectral sequence degeneration. These issues are local and fixable, so I recommend major revision rather than rejection. The sign error in Lemma 2.8 is easily corrected and does not affect the central claim."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The genuinely new result is Theorem 2.9/2.10: for a special Gushel-Mukai threefold, the anti-invariant deformations exhaust the 3-dimensional kernel of dP, and the invariant part of the infinitesimal period map is injective. That is the right analog of the hyperelliptic curve picture, and it is not in Debarre–Iliev–Manivel. The paper also computes kernel dimensions for prime Fano threefolds g=7,8,9,10,12 and gives a Bridgeland interpretation of the kernel for g=6,8. The categorical machinery is substantial and, as far as I can tell, correctly deployed; reducing special GM to ordinary GM via [KP23] and using normal Hochschild cohomology is legitimate, and the cited results are real.\n\nThe paper's soft spot is the Hodge proof at Lemma 2.17. The reduction of Theorem 2.10 to the K3 pairing (Proposition 2.16) passes through a commutative diagram that the authors do not prove; they say it follows from exact sequences (19), (20) and [Fle86, Lemma 2.10]. The vertical maps are composites of contraction, residue, and boundary maps, so sign or twist errors are not cosmetic. If Lemma 2.17 fails, Lemmas 2.18 and 2.21 do not imply Proposition 2.16. The stress-test note is right about this.\n\nMinor issues are more irritating than damaging. Lemma 2.8's sign argument is written inconsistently: the cup product with differential forms is graded-commutative, so the displayed contradiction needs care; the conclusion is almost certainly correct via Proposition 2.2. Lemma 2.19 (restriction H^1(Y,Ω^1_Y)→H^1(S,Ω^1_Y|_S) is an isomorphism) is stated with a proof that looks right. Lemma 2.17, however, is not a minor omission; it is the hinge of the Hodge argument. The authors should supply the missing diagram chase or clearly mark that step as a conjecture to be checked.\n\nI want to stress that the categorical proof is independent, so the main theorem does not collapse without Lemma 2.17. The paper is not circular: special GM is reduced to ordinary GM via published equivalences, and the Hodge part is benchmarked against Flenner and Debarre–Iliev–Manivel. The kernel descriptions for g=7–12 are a real byproduct.\n\nWho should read this: anyone working on Torelli problems for Fano threefolds or on Kuznetsov components of GM varieties. It deserves a serious referee. I would send it out and ask that the Hodge gap be closed or explicitly flagged as deferred.","headline":"The main theorem on special Gushel-Mukai threefolds is new and plausible; the Hodge proof has one unproved, load-bearing commutativity lemma, but the independent categorical argument makes this a serious paper.","tokens_in":30289,"tokens_out":1944,"would_cite":true,"duration_ms":21740,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["14F08","14J45","14J10","14C34"],"pacs":[],"model":"deepseek-v4-flash","headline":"For a special Gushel-Mukai threefold, the infinitesimal period map has kernel dimension 3, consisting exactly of the deformations that move the double cover to an ordinary Gushel-Mukai threefold.","keywords":["Bridgeland stability","Gushel-Mukai threefold","Hochschild homology","Kuznetsov component","Torelli problem","infinitesimal period map","K3 surface","normal Hochschild cohomology"],"falsifier":"Compute $\\dim H^{1}(S,\\Omega^{1}_Y|_S)$ for a smooth anticanonical K3 surface $S=Y\\cap Q$ with $Y=\\mathrm{Gr}(2,5)\\cap\\mathbb{P}^{6}$: Lemma 2.19 predicts it is 1, and any other value would invalidate the proof of injectivity of $I$ and hence of the invariant period map. Equivalently, evaluate the pairing $H^{1}(S,T_S)_{0}\\otimes H^{1}(S,\\Omega^{1}_S(-1))\\to H^{2}(S,\\mathcal{O}_S(-1))$ on an explicit special Gushel-Mukai threefold and look for a non-zero first-factor element that pairs to zero with every class in $H^{1}(S,\\Omega^{1}_S(-1))$; its image in $\\mathrm{Ker}\\,dP$ would disprove Theorem 2.9.","tokens_in":29261,"feed_emoji":"📐","tokens_out":16151,"duration_ms":126619,"temperature":0.7,"pith_summary":"The paper proves that a special Gushel-Mukai threefold—a double cover of the rigid Fano threefold $Y=\\mathrm{Gr}(2,5)\\cap\\mathbb{P}^{6}$ branched along a K3 surface $S$—behaves like a hyperelliptic curve for the infinitesimal Torelli problem: the only deformations invisible to the period map are the three that deform the double cover into an ordinary Gushel-Mukai threefold. Equivalently, the invariant part of the infinitesimal period map, which records how the branch K3 moves inside $Y$, is injective. The proof reduces this to a twisted infinitesimal Torelli statement for $S$, and a parallel categorical proof, working with the Kuznetsov component of the derived category, gives the same kernel computation. The same methods compute the kernel of the period map for prime Fano threefolds of genus 7, 8, 9, 10, and 12, and for special Verra threefolds, and identify the kernel geometrically with the tangent space of a Bridgeland moduli space of stable objects.","feed_headline":"Special Gushel-Mukai period map: kernel is 3-dimensional","feed_subtitle":"Only the directions that turn the double cover into an ordinary Gushel-Mukai threefold are invisible.","key_machinery":"The central mechanism is a reduction of the invariant period map to a twisted cup product on the branch K3 surface. The invariant deformation space $H^{1}(Y,T_Y(-\\log S))$ injects into $H^{1}(S,T_S)$ as the image $H^{1}(S,T_S)_{0}$ of the Kodaira–Spencer map $H^{0}(S,N_{S|Y})\\to H^{1}(S,T_S)$, and residue sequences for logarithmic forms set up a commutative diagram up to sign that identifies the invariant pairing with $H^{1}(S,T_S)_{0}\\otimes H^{1}(S,\\Omega^{1}_S(-1))\\to H^{2}(S,\\mathcal{O}_S(-1))$. The load-bearing identity is the injectivity of the connection map $I$ in Lemma 2.18, which rests on Lemma 2.19: the restriction map $H^{1}(Y,\\Omega^{1}_Y)\\to H^{1}(S,\\Omega^{1}_Y|_S)$ is an isomorphism, so $\\dim H^{1}(S,\\Omega^{1}_Y|_S)=1$, forcing certain classes to be multiples of the polarization and producing the contradiction that proves injectivity. On the categorical side the argument runs through the Kuznetsov component $\\mathrm{Ku}(X)$ (the nontrivial factor of the derived category after removing exceptional line bundles), the Hochschild-cohomology action map $\\gamma_X: HH^{2}(\\mathrm{Ku}(X))\\to \\mathrm{Hom}(HH^{-1}(\\mathrm{Ku}(X)), HH^{1}(\\mathrm{Ku}(X)))$, normal Hochschild (co)homology, and the Bridgeland moduli space $M_{\\sigma}(\\mathrm{Ku}(X),[I_C])$, whose tangent space at the distinguished point is $\\mathrm{Hom}(U_X,Q^{\\vee}_X)$.","core_discovery":"Let $X$ be a special Gushel-Mukai threefold, the double cover of $Y=\\mathrm{Gr}(2,5)\\cap\\mathbb{P}^{6}$ branched along a degree-10 K3 surface $S\\in|-K_Y|$. The main theorem (Theorem 2.9) asserts that the kernel of the infinitesimal period map $dP: H^{1}(X,T_X)\\to \\mathrm{Hom}(H^{1}(X,\\Omega^{2}_X), H^{2}(X,\\Omega^{1}_X))$ is exactly the anti-invariant summand $H^{1}(Y,T_Y(-1))$, of dimension 3. By the decomposition of $dP$ into invariant and anti-invariant parts under the covering involution (Proposition 2.2), this is equivalent to the injectivity of the invariant map $H^{1}(Y,T_Y(-\\log S))\\to \\mathrm{Hom}(H^{1}(Y,\\Omega^{2}_Y(\\log S)(-1)), H^{2}(Y,\\Omega^{1}_Y(\\log S)(-1)))$. The Hodge-theoretic proof reduces this to a twisted Torelli statement for the branch K3: the pairing $H^{1}(S,T_S)_{0}\\otimes H^{1}(S,\\Omega^{1}_S(-1))\\to H^{2}(S,\\mathcal{O}_S(-1))$ is non-degenerate on the codimension-one subspace $H^{1}(S,T_S)_{0}$ that records deformations of $S$ inside $Y$. The categorical proof shows that the kernel of the categorical period map $\\eta: H^{1}(X,T_X)\\to HH^{2}(\\mathrm{Ku}(X))$ equals the classical kernel, and for prime Fano threefolds of genus $g\\ge 6$ it is $\\mathrm{Hom}(U_X,Q^{\\vee}_X)$ for $g=6,8$, a quotient for $g=7,9,10$, and $H^{1}(X,T_X)$ for $g=12$.","pith_inferences":["If the main theorem holds, the 3-dimensional kernel should be readable as the tangent space of the period-partner locus inside the ordinary Gushel-Mukai moduli space; the connection the authors ask about in Question 1.7 would then be a direct geometric isomorphism $H^{1}(Y,T_Y(-1))\\cong \\mathrm{Hom}(U_X,Q^{\\vee}_X)$, rather than the chain of equalities used here.","The Hodge-theoretic strategy should generalize to any double cover of a rigid Fano threefold branched along an anticanonical K3: the kernel dimension should equal $h^{1}(Y,T_Y\\otimes L^{-1})$, computable from the same twisted-K3 pairing, making special Verra threefolds one instance of a uniform pattern.","Because the categorical argument makes the kernel depend only on the Kuznetsov component, the ordinary-versus-special difference in kernel dimension (2 versus 3 for Gushel-Mukai threefolds) should be visible as a change in the distinguished object of the component, not in the component itself; testing this on the categorical duality would be a concrete check.","A direct numerical computation of the twisted pairing on one explicit special Gushel-Mukai K3 would test the non-degeneracy claim independently of the vanishing theorems; if a null vector within $H^{1}(S,T_S)_{0}$ appeared, the Hodge-theoretic proof would need a revision in Lemma 2.18."],"forward_implications":["For a special Gushel-Mukai threefold, the kernel of $dP$ is the 3-dimensional anti-invariant space $H^{1}(Y,T_Y(-1))$: the two directions already invisible for ordinary Gushel-Mukai threefolds together with the one direction that turns the double cover into an ordinary one.","The invariant part of the period map is injective: no first-order deformation that keeps the threefold special, i.e. a deformation of the branch K3 inside $Y$, is invisible to period data.","For prime Fano threefolds the kernel dimensions are: genus 7, zero (infinitesimal Torelli holds); genus 8, 5; genus 9, 6; genus 10, 7; genus 12, 6 (Corollary 3.3).","Special Verra threefolds have 1-dimensional period-map kernel, while ordinary Verra threefolds satisfy the infinitesimal Torelli theorem (Propositions 1.3 and A.2).","For genus 6 and 8 threefolds, $\\mathbb{P}\\mathrm{Ker}\\,dP\\cong \\mathbb{P}\\mathrm{Hom}(U_X,Q^{\\vee}_X)$ is realized geometrically as the exceptional locus (or divisor) of the birational morphism from the Hilbert scheme of conics, or the moduli space of semistable sheaves, to the Bridgeland moduli space (Theorem 1.5)."],"supporting_citations":[{"why":"Supplies the decomposition of the infinitesimal period map into invariant and anti-invariant parts for a double cover, which sets up the whole Hodge-theoretic reduction.","marker":"[Kon85]"},{"why":"Computes the 2-dimensional kernel for ordinary Gushel-Mukai threefolds, the baseline the special case is compared with, and provides the period-map fiber description used later.","marker":"[DIM12]"},{"why":"Supplies the infinitesimal Torelli techniques and the lemmas used to prove non-degeneracy of the twisted pairing for the branch K3 surface.","marker":"[Fle86]"},{"why":"Categorical duality: a special Gushel-Mukai threefold shares a Kuznetsov component with an ordinary one, transferring injectivity of the categorical period map to the special case.","marker":"[KP23]"},{"why":"Establishes the diagram relating the classical and categorical infinitesimal period maps and proves injectivity of the categorical map in the ordinary Gushel-Mukai case.","marker":"[JLLZ23]"},{"why":"Constructs the birational morphism from the Hilbert scheme of conics to the Bridgeland moduli space and identifies the tangent space at the distinguished point with Hom(UX, Q∨X).","marker":"[JLLZ24]"},{"why":"Introduces normal Hochschild (co)homology and its spectral sequence, which give the computation of Ker η.","marker":"[Kuz15]"},{"why":"Reconstruction theorem used to identify the point of the Bridgeland moduli space that corresponds to the isomorphism class of X, needed for the geometric description of the kernel.","marker":"[Log12]"}],"fun_headline_variants":["Special Gushel-Mukai Torelli: invariant part injective, kernel 3D","Invariant period map injective on special Gushel-Mukai","Kernel of GM period map is 3D, from anti-invariant part","Infinitesimal Torelli for special GM: only anti-invariant deformations invisible"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The argument for the Hodge-theoretic proof rests on Lemma 2.19, which claims that the restriction map $H^{1}(Y,\\Omega^{1}_Y)\\to H^{1}(S,\\Omega^{1}_Y|_S)$ is an isomorphism and hence that $\\dim H^{1}(S,\\Omega^{1}_Y|_S)=1$; if that cohomology computation fails, the non-degeneracy of the twisted pairing in Proposition 2.16 would be unsupported and the invariant part of the period map could have additional kernel.","fun_headline_variants_meta":{"raw":{"variants":["Special Gushel-Mukai Torelli: invariant part injective, kernel 3D","Invariant period map injective on special Gushel-Mukai","Kernel of GM period map is 3D, from anti-invariant part","Infinitesimal Torelli for special GM: only anti-invariant deformations invisible"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001192,"raw_usage":{"total_tokens":5075,"prompt_tokens":1259,"completion_tokens":3816,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":875,"completion_tokens_details":{"reasoning_tokens":3727}},"tokens_in":875,"tokens_out":3816,"duration_ms":40542,"temperature":1.0,"reasoning_tokens":3727,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-06T18:52:17.069027+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compute $\\dim H^{1}(S,\\Omega^{1}_Y|_S)$ for a smooth anticanonical K3 surface $S=Y\\cap Q$ with $Y=\\mathrm{Gr}(2,5)\\cap\\mathbb{P}^{6}$: Lemma 2.19 predicts it is 1, and any other value would invalidate the proof of injectivity of $I$ and hence of the invariant period map. Equivalently, evaluate the pairing $H^{1}(S,T_S)_{0}\\otimes H^{1}(S,\\Omega^{1}_S(-1))\\to H^{2}(S,\\mathcal{O}_S(-1))$ on an explicit special Gushel-Mukai threefold and look for a non-zero first-factor element that pairs to zero with every class in $H^{1}(S,\\Omega^{1}_S(-1))$; its image in $\\mathrm{Ker}\\,dP$ would disprove Theorem 2.9.","supporting_citations":[],"review_version":1}